| The "Montserrat-2000" severe flash flood event which occurred
over Catalonia on 9 and 10 June 2000 is analyzed.
Strong precipitation was generated by a mesoscale convective system
associated with the development of a cyclone.
The location of heavy precipitation depends
on the position of the cyclone, which, in turn, is found to
be very sensitive to various model characteristics and initial
conditions.
Numerical simulations of this case study using
the hydrostatic BOLAM and the non-hydrostatic MOLOCH models
are performed in order to test the effects of different formulations
of the boundary layer parameterization:
a modified version of the Louis (order 1) model
and a custom version of the E-ℓ (order 1.5) model.
Both of them require a diagnostic formulation of the mixing length,
but the use of the turbulent kinetic energy equation
in the E-ℓ model
allows to represent turbulence history and non-locality
effects and to formulate a more physically based mixing length.
The impact of the two schemes is different in the two models.
The hydrostatic model, run at 1/5 degree resolution, is less sensitive,
but the quantitative precipitation forecast is in any case
unsatisfactory in terms of localization and amount.
Conversely, the non-hydrostatic model, run at 1/50 degree resolution,
is capable of realistically simulate timing, position and amount
of precipitation, with the apparently superior results obtained
with the E-ℓ parameterization model. |